20 . VARIATION IN ASTER 
impossible to say from how many seedlings a whole connected 
field full of A. dumosus, A. paniculatus or A. ericoides may have 
developed. 
In attempting to determine how far the plants of a single 
colony may vary from each other, I have made observations on 
a series of widely scattered examples selected with reference to 
continued opportunity of access, for periods varying from 3 to 8 
years, chiefly the latter, and including the following localities : 
Washington, D. C., on the Potomac, along Rock Creek, at 
Terra, Cotta, etc, 
New York city, in Central Park ; at Inwood; on the Palisades ; 
in Van Cortlandt Park ; at Bryn Mawr Park ; at Hillview, Mosh- 
olu, Neperhan, Grassy Sprain Lake, etc. 
Western New York, in the Lake Erie region, in the Cattarau- 
gus Indian reservation ; in Silver Creek ravine; at Swift’s Hill, 
Silver Creek ; at Talcott’s Woods, in Sheridan; at Point Gratiot 
in Dunkirk ; at Fredonia; in Niagara gorge; etc. 
White Mountain region, about the Franconia Notch, the An- 
droscoggin and Peabody Rivers, etc. 
Taconic Mountain region, Mt. Washington, Mass. 
Massachusetts interior, Charles River in Weston. 
Martha’s Vineyard, at numerous localities. 
These continued comparisons, together with less prolonged 
observation at many other places, give evidence that, in Aster, 
sports are apparently very common among seedlings, showing ab- 
rupt divergence from type; and that gradual modification on the 
other hand is very common among root-propagated plants, show- 
ing rapid but not very abrupt adaptations to environment and 
usually departing from the type rather in degree than in kind. 
For example, in one large plantation composed wholly of 
thousands of plants of <A. quiescens (watched for years on the 
Palisades but now unhappily destroyed) while perhaps nine out of 
ten plants conformed quite strictly to type, the following adaptive 
modifications occurred, producing what might at first have been 
taken in some cases for new species. 
. Stump-fed plants, with great increase of folios and de- 
crease of inflorescence where nourished from a well-rotted oak- 
stump ; the naturally large basal leaves being somewhat larger, 
